What a Smartphone “Platform” Really Is (Boundary + Role Map)

A smartphone platform is a tightly coupled system where compute, RF, power, human interfaces, and sensors share the same battery, ground, thermal budget, and mechanical stack. The goal of this chapter is to turn common user symptoms into a first-pass hardware attribution path with measurable evidence.

• In-scope: device-side blocks and coupling paths (power droop/ripple, ground return, thermal derating, EMI/ESD impacts, measurable logs).
• Out-of-scope: OS/app/cloud architecture, operator/network-side optimizations, and certification walkthroughs.

• Compute (AP/Baseband + LPDDR/UFS) creates fast load steps that stress rails and heat paths.
• RF (RFIC + FEM/PA/LNA/filters + antennas) is sensitive to PA supply droop, temperature, and broadband noise.
• Power (VBAT → PMIC rails) is the shared constraint; rail impedance peaks translate into resets, link drops, and interface noise.
• HMI (Display/Touch/Audio) is a noise “detector”: ground bounce and switching noise appear as flicker, ghost touch, pop/noise.
• Sensors (IMU/ALS/PS/Baro/Fingerprint/ToF) drift or false-trigger under supply noise, thermal gradients, and mechanical stress.

• Random reboot / freeze
  First check: VBAT droop + key rail dip → PMIC fault/IRQ + reset reason → thermal throttle events. (Power/Compute)
• Throughput oscillation / call drop
  First check: PA rail droop during Tx burst → power backoff/derating events → antenna tuning state + RSSI/AGC trend. (RF + Power + Thermal)
• Ghost touch / touch drift
  First check: touch raw noise map + display supply ripple → ground return loops near touch/display → post-ESD behavior delta. (HMI + ESD)
• Audio pop / hiss / periodic buzz
  First check: audio FFT noise lines (switching harmonics) → mic bias ripple/ground loop → class-D return path proximity. (Audio + Power/Return)
• Sensor “looks like algorithm” but feels wrong
  First check: sensor self-test/cal data → drift vs temperature scan → supply noise vs sensor sampling window. (Sensors + Thermal/Power)

Next step: treat compute and modem as state machines that create load steps and heat. Chapter H2-2 formalizes the measurement-first method to align symptoms with the earliest electrical or thermal event.

Apps Processor + Baseband: Data Paths That Drive Power & Heat

CPU/GPU/NPU and the modem are not only “compute blocks”—they are primary load-step generators. Their state transitions (idle → burst → sustained) reshape rail impedance stress, VBAT droop, and thermal headroom, which then cascades into RF derating, UI noise, and stability events.

• Idle: low baseline; noise is often dominated by switching converters and display refresh.
• Burst (boost): fast current steps; exposes rail impedance peaks and weak decoupling/return loops.
• Sustained load: heat accumulation; triggers derating (thermal throttling, RF power backoff).

• LPDDR activity can create fast edge-correlated current steps; weakness appears as core/memory rail ripple and ground bounce.
• UFS bursts (write spikes / background maintenance) can correlate with VBAT dips and EMI-sensitive interface upsets.
• Modem state can trigger RF Tx bursts; PA supply droop and temperature rise become the dominant constraints.

• Step A — Align time
  Capture VBAT + one key rail (core or PA) + temperature with a shared timestamp. Symptoms must be correlated to transitions, not averages.
• Step B — Read the platform’s “truth” logs
  Collect PMIC fault/IRQ flags and reset reason registers. A “random reboot” without reset cause is not evidence yet.
• Step C — Separate droop vs derating
  Droop: immediate voltage dip around the transition. Derating: delayed performance drop tied to temperature or protection thresholds.
• Step D — Link to user-visible metrics
  RF: power backoff/throughput oscillation. HMI: touch noise map changes. Audio: switching-harmonic lines in FFT.

• Power integrity: reduce rail impedance peaks (decoupling placement, loop area, return path continuity), prioritize rails that correlate to the symptom timing.
• Domain isolation: keep noisy switching returns away from touch/audio references; avoid shared bottleneck vias/planes on sensitive paths.
• Thermal headroom: improve heat flow from hot spots (AP, PA, PMIC) to spreaders; ensure temperature sensing is representative of the true hotspot.

Next chapters (not included in this delivery) will expand the power-domain view (PMIC rails, sequencing, protection logs) and then connect those findings to RF stability, HMI noise windows, and sensor drift—without crossing into OS/app/network-side topics.

PMIC & Power Domains: Rails, Sequencing, and Why “Random Reboot” Happens

“Random reboot” is rarely random. Most cases collapse into a small set of device-side mechanisms: VBAT impedance and droop, a specific rail dip, a sequencing dependency violation, or a protection event that triggers earlier than expected. This chapter focuses on measurable proof—waveforms and PMIC/reset logs— before design changes are attempted.

• In-scope: VBAT → PMIC → rails, sequencing/dependencies, UVLO/OCP/thermal/brownout behavior, PMIC IRQ/fault logs, reset reason registers.
• Out-of-scope: adapter AC-DC topology (PFC/flyback/LLC), network/operator-side causes, OS/app/cloud architecture.

• Core domain: highest load-step source; dips often map to immediate reset or hard freeze.
• Memory/IO domain: noise-sensitive; dips can cause silent corruption, bus faults, or delayed watchdog reset.
• RF/PA domain: sensitive to droop and heat; often shows link instability before any reboot.
• Display/Touch domain: a “noise detector”; ground bounce/ripple surfaces as flicker, ghost touch, or controller resets.
• Audio domain: reference-ground sensitive; ripple/return-loop issues show as pop/hiss/buzz and codec brownouts.
• Sensor domain: low-power but drift-prone; supply noise can cause false triggers that cascade into higher system load.

• Chain reset (direct)
  A core-domain dip crosses an internal threshold → PMIC flags UVLO/brownout → reset asserted → platform restarts.
• Delayed reset (indirect)
  A non-core domain dips (memory/display/audio) → system hangs or misbehaves → watchdog/reset reason indicates delayed recovery.
• Dependency violation
  One rail comes up/down out of order (or collapses faster) → dependent block faults → reset reason may not look like “power” unless logs are captured correctly.

• Brownout / UVLO “invisible dip”
  A short droop is caught by PMIC comparators, but a distant probe or low bandwidth measurement misses it. Evidence must be taken close to PMIC input/rail with short ground.
• OCP “current not high” illusion
  Fast di/dt and parasitic inductance produce a spike at the sense point; protection trips even when average current looks normal. Correlate the trip to the transition edge.
• Thermal “average is fine” illusion
  A hotspot (PA, PMIC, or SoC) crosses limit while a remote sensor reads acceptable. Correlate derating/thermal flags with localized temperature evidence.

• Step A — Pick the rail based on symptom timing
  Immediate reboot: start with VBAT + core rail. Link instability first: start with PA rail. UI noise first: start with display/touch rail.
• Step B — Capture the transition edge
  Trigger on dip/ripple around the state change (boost, Tx burst, camera start). Averages hide the event that trips thresholds.
• Step C — Read PMIC truth signals
  Collect PMIC IRQ/fault bits and timestamp them relative to the waveform. A “trip class” (UVLO/OCP/thermal) narrows the fix immediately.
• Step D — Confirm with reset reason
  POR/BOR/WDOG/thermal reset codes separate “power collapse” from “delayed hang recovery.” Use this to avoid mislabeling software as the primary cause.

• VBAT path: reduce input impedance (connector/trace inductance, local bulk placement near PMIC input) if droop is captured on VBAT.
• Critical rail PDN: flatten impedance peaks (decoupling mix and placement, return-path continuity) if rail dips correlate with state transitions.
• Protection robustness: tune filtering/blanking and ensure sense points reflect real hotspots/loads if mis-trips are observed in PMIC logs.

Next chapter ties the same reboot evidence to the physical causes—decoupling tiers, return paths, and coupling channels that turn a rail into a symptom.

Transient Load, Decoupling, and Ground Return: Where PI/SI/EMI Intersect

When specifications look “correct” but experience is poor, the root cause is often physical: impedance peaks, loop area, and return-path bottlenecks that push noise into sensitive references. In a smartphone, PI, SI, and EMI are not separate problems—they are different views of the same current loops.

• Transient load: AP boost and RF Tx bursts generate fast current steps with broad frequency content.
• PDN response: any impedance peak turns that step into rail dip/ripple and ground bounce.
• Symptom mapping: a dip may cause reset; ripple/ground bounce may cause RF instability, touch noise, or audio buzz—without a reboot.

• Bulk (energy reservoir)
  Targets VBAT droop and lower-frequency sag. Weak bulk placement often shows as brownout-like resets under load steps.
• Mid (impedance shaping)
  Fills the impedance “valley” between bulk and high-frequency decaps. Weak mid-tier often shows as rail ripple that correlates with RF/UI events.
• High-frequency (local loop closure)
  Closes the fastest current loops at the load pins. Weak HF loop closure often shows as ground bounce, EMI hot spots, and interface upsets.

• Power coupling: shared rails or shared input path spreads ripple and droop between domains.
• Ground coupling: overlapping returns and via/plane bottlenecks create ground bounce that shifts “reference” for touch/audio/RF.
• Trace coupling: switching nodes and fast edges capacitively/inductively inject noise into sensitive routes and connectors.

• Step A — Find EMI hot spots (near-field scan)
  Scan around PMIC switch nodes, PA supply path, display/backlight area, and touch FPC region. Mark the strongest harmonics.
• Step B — Measure rail ripple where it matters
  Probe close to the load (core/PA/display/audio reference). Long ground leads can hide the real ripple and invent false ringing.
• Step C — Correlate in time
  Align ripple/EMI bursts with system transitions (boost, Tx burst, brightness change, touch scan window). Correlation beats speculation.

• Decap placement: HF decaps must minimize loop area to the exact load pins; mid-tier must avoid long stubs.
• Return vias: avoid single-via bottlenecks on high di/dt paths; provide parallel returns where current is pulsed.
• Switch node containment: keep switching nodes short and shielded from touch/audio references and sensitive connectors.
• Domain boundary: keep RF and HMI reference grounds clean; prevent large power returns from crossing those references.
• Connector regions: ensure ESD/edge noise has a short, direct return to chassis/ground reference, not through signal reference paths.
• FPC routing: avoid running touch/display flex routes over power hot loops; maintain consistent return adjacency.

• If reboot correlates to droop: strengthen VBAT bulk near PMIC input and flatten PDN peaks on the affected rail.
• If RF instability correlates to Tx burst sag: prioritize PA rail loop closure and reduce shared impedance in the input path.
• If touch/audio noise correlates to switching harmonics: isolate returns, reduce coupling from switch nodes, and relocate sensitive references away from hot loops.

RF Tx/Rx + PA/FEM: Throughput Drops, Call Drops, and Thermal Derating

“Bad signal” on a smartphone is rarely a single cause. On the device side, most field issues fall into a small set of measurable buckets: genuine weak coverage, receiver blocking/intermodulation, antenna mismatch and tuner state problems, PA supply sag during Tx bursts, or thermal derating that forces power backoff. This chapter focuses on evidence that can be time-aligned to throughput/call events.

• In-scope: RFIC, PA/FEM, filters/switches, antenna/tuner, diversity, PA supply, device thermal derating evidence.
• Out-of-scope: base-station/RAN topics, operator-side optimization, network architecture.

• A) True weak signal (link budget)
  RSRP/RSSI are consistently low and AGC is near its limit. Degradation is steady, not bursty.
• B) Blocking / intermod / self-interference
  RSRP may look acceptable, but noise floor rises, AGC behavior is abnormal, and throughput collapses under certain environments or bands.
• C) Antenna mismatch / grip / tuner state
  Performance changes strongly with orientation/grip; tuner/diversity state transitions correlate with throughput swings.
• D) PA supply sag on Tx bursts
  Short, repeatable drops align with Tx activity; PA power backoff events appear near the same timestamps.
• E) Thermal derating
  Performance starts strong then decays as temperature rises; power backoff/derating events are temperature-correlated.

• Tx bursts are fast current steps at the PA/FEM supply, often more decisive than average dissipation.
• Supply sag/ripple can force power backoff, raise EVM/ACLR risk, and increase BLER—felt as throughput drops or call instability.
• Key trap: “VBAT looks fine” does not prove “PA rail is fine.” The highest value measurement is near the PA supply node.

• Diversity/MIMO changes effective SNR margin and blockage resilience; the benefit is often seen as stability rather than peak speed.
• Antenna tuning modifies match/efficiency; failed or late state transitions can produce oscillating throughput patterns.
• Evidence-first: treat tuner state, diversity selection, and band/Tx mode as first-class “events” to align with throughput drops.

• Step A — Decide “weak signal” vs “not weak signal”
  Use RSSI/RSRP plus AGC trend. True weak signal typically shows persistent low levels and AGC near the edge.
• Step B — Check power backoff / derating events
  Look for PA power reduction flags or backoff counters near the drop timestamp; these separate RF chain constraints from pure coverage.
• Step C — Correlate with temperature
  If backoff grows with temperature, thermal derating is the likely primary constraint. If it appears on bursts at any temperature, suspect supply sag or mismatch.
• Step D — Confirm with a targeted rail check (if available)
  Measure PA rail sag/ripple close to the supply node and align it with Tx burst timing and backoff events.

Next chapter shifts to the audio path, where many “codec quality” complaints are actually power/return-path and transient-control problems.

Audio Subsystem: Codec, Amps, Mics—Pop/Noise/ANC Failure Modes and Proof

Many smartphone audio failures are diagnosed as “codec quality,” but field evidence often points to power integrity, return paths, bias stability, and transient control. This chapter translates subjective symptoms (hiss, buzz, pop, echo, ANC instability) into measurable signatures (FFT spectrum, time-domain transients, ground/return checks, and codec/amp diagnostic registers).

• In-scope: mic bias and mic front-end path, codec, headphone amp, speaker class-D amp, jack detect/switching, phone-side EMI injection and return paths.
• Out-of-scope: TWS/charging case, external speakers/soundbar system design.

• Mic / Record path
  Mic → mic bias → input AFE / codec ADC → DSP. Sensitive to bias noise, reference ground shifts, and EMI injection near flex/connectors.
• Playback path
  codec DAC → headphone amp or speaker class-D → load. Sensitive to pop suppression sequencing, large return currents, and switching harmonics.
• Control path
  jack detect / route switching / gain ramps. Many “pop” cases are event-chain bugs (timing + bias settling), not codec limitations.

• Hiss / buzz / hum
  FFT shows tonal components or harmonics near switching frequencies; often points to rail ripple, ground return coupling, or insufficient filtering on bias/reference nodes.
• Pop on plug/unplug or route switch
  Time-domain transient spike aligns with detect/switch events. Bias settling and enable sequencing are the first suspects.
• Recording distortion / clipping
  FFT shows harmonics and flattening; can be caused by mic bias instability, front-end overload, or reference ground bounce during system events.
• ANC unstable / ineffective
  Sensitivity mismatch or injected noise on mic paths reduces coherence. Evidence is seen as inconsistent spectra across mics and event-tied disturbances.

• Mic bias integrity: bias ripple and reference shifts enter the ADC as “audio.” Bias filtering and return-path cleanliness matter more than raw codec specs.
• Class-D return currents: speaker amps move large pulsed currents; if those returns share sensitive references, noise appears as buzz, pops, or mic contamination.

• Event sequence
  Detect → route switch → bias settle → gain ramp → amp enable. Pops appear when any step is too fast or out of order.
• Proof points
  Capture transient at output/bias nodes; read codec/amp diagnostic registers; align timestamps to the detect/switch event.

• Step A — FFT classification
  Identify tonal switching harmonics vs wideband noise vs clipping distortion. Classification determines where to probe next.
• Step B — Transient capture
  Record plug/unplug and route-switch events; measure output and bias nodes. Pops are time-domain problems first.
• Step C — Return path / ground loop check
  Verify that class-D high-current returns do not traverse mic/codec reference grounds. Look for bottlenecks and shared vias near sensitive nodes.
• Step D — Codec/amp diagnostic registers
  Read undervoltage/overcurrent/thermal flags, gain/state bits, pop-suppression state, and route status. Registers provide “why” when waveforms show “when.”

Display & Touch: MIPI DSI + Backlight/PMIC + Touch Controller Coupling

Screen artifacts and touch anomalies often come from coupling between display interface timing, backlight switching, touch scan windows, and shared power/ground paths. Typical field complaints—flicker, random lines, ghost touch, and touch drift after ESD— can be classified by measurable signatures: touch raw data/noise maps, rail ripple near the load, and before/after drift comparisons.

• In-scope: MIPI DSI link behavior as evidence, display/backlight/touch rails, scan-window coupling, grounding/shielding/FPC effects, ESD-after drift.
• Out-of-scope: TV/monitor scaler and deep TCON architecture topics.

• Power noise
  Display rail, backlight rail, and touch rail ripple/steps can modulate pixels and shift touch baselines.
• Ground return / shielding gaps
  Shared return paths and imperfect shielding around FPC/connectors can inject switching and ESD energy into sensitive references.
• Timing windows
  Touch scans occur in windows; interference becomes visible when backlight PWM or mode changes overlap those windows.
• ESD-after parameter drift
  The panel may keep working, while touch baseline/thresholds drift—showing up as ghost touches or instability.
• Connector/FPC intermittency
  Mechanical stress or contact resistance changes can create intermittent lines, flicker, or event-triggered touch faults.

• Flicker / brief artifacts
  Evidence: rail ripple increases during brightness/refresh transitions; event timing matches. Misread: “panel defect” without correlation checks.
• Random lines / intermittent green line
  Evidence: sensitivity to flex/temperature; connector/FPC intermittency patterns. Misread: treating every line as a permanent panel failure.
• Ghost touch / touch drift
  Evidence: raw data baseline shifts; noise map hotspots; strong dependence on PWM/charging/near-field noise. Misread: changing only software sensitivity.
• After-ESD touch abnormal
  Evidence: before/after baseline and drift delta; controller status changes; increased false-trigger rate. Misread: “ESD passed so it cannot be ESD.”

• Touch is windowed: scans and integrations occur in defined windows, not continuous “always-on” sampling.
• Coupling becomes visible when PWM edges or mode transitions overlap scan windows, creating repeatable ghost-touch patterns.
• Proof pattern: touch raw/noise maps improve when scan timing is shifted or when the interfering event (PWM/mode switch) is altered.

• Before/after delta is the signal: baseline shift, threshold drift, and false-trigger statistics matter more than “alive/dead.”
• Key comparison: raw data baseline + noise map + drift magnitude across the same test pattern, pre- and post-ESD.

• Step A — Capture touch raw + noise map
  Identify baseline shift and noise hotspots; compare across brightness levels and mode transitions.
• Step B — Probe rails near the load
  Measure display/backlight/touch rails ripple and steps; correlate with artifact/touch events.
• Step C — Time-align with scan windows
  Check whether PWM edges or refresh/mode switches overlap scan windows; validate by shifting timing or changing duty/frequency.
• Step D — ESD before/after comparison
  Repeat the same raw/noise-map acquisition; quantify drift delta and false-trigger rate changes.

Rich Sensor Suite: IMU/ALS/PS/Baro/Mag/Fingerprint/ToF—Drift and False Triggers

“Step count is off,” “raise-to-wake fails,” “fingerprint stops working,” and “altitude jumps” usually trace back to a small set of hardware-dominant causes: calibration integrity, temperature-driven drift, mechanical stress/assembly effects, or power/bus noise that creates false triggers. This chapter focuses on proof artifacts—self-test and calibration data, temperature sweeps, and noise density/bias drift curves.

• In-scope: sensor rails and isolation, I²C/SPI reliability as evidence, self-test and calibration records, temperature sweeps, bias drift and noise density behavior, mechanical stress effects.
• Out-of-scope: medical-grade algorithms, health scoring and clinical inference.

• Inertial (IMU)
  Bias drift and noise density determine stability; stress and temperature shifts can look like “algorithm issues.”
• Optical (ALS/PS) and windows
  Ambient light and proximity can false-trigger due to optical path changes, contamination, or rail/bus noise.
• Baro (altitude)
  Temperature gradients, sealing/port effects, and rail noise create step-like altitude jumps and drift.
• Mag (compass)
  Assembly shifts and nearby magnetic materials change offsets; temperature and stress can add drift.
• Fingerprint + ToF
  ESD sensitivity, reference stability, window alignment and optical coupling determine success rate and false rejects.

• A) Calibration integrity
  Calibration missing, overwritten, or inconsistent after events; self-test results and stored calibration parameters expose this quickly.
• B) Temperature-driven drift
  Bias and sensitivity drift across temperature; the signature is a repeatable bias-vs-temp curve with clear slopes or knees.
• C) Mechanical stress / assembly
  Offsets jump with press/flex or after thermal cycling; IMU/mag and optical alignments can shift with adhesive, torque, or housing deformation.
• D) Power or bus noise
  Noise density increases or data becomes bursty; I²C/SPI error/retry counts and timing anomalies align with false triggers.

• Sensor rail isolation prevents high di/dt domains from modulating biases and references.
• Bus integrity is evidence: error/retry bursts and abnormal timing often align with false triggers more strongly than “raw readings.”
• Best practice for proof: compare noise density and error counters across system states (bright display, charging, radio active).

• IMU/mag: stress can translate into bias/offset changes; a press/flex test often reveals step-like offset jumps.
• ALS/PS/ToF/fingerprint: window alignment, contamination, and mechanical stack-ups create false triggers and failure clusters.
• Proof pattern: compare calibration/self-test and offset distributions before/after assembly operations and thermal cycling.

• Step A — Read self-test and calibration records
  Capture self-test pass/fail and calibration parameters; repeat before and after key events (reset/ESD/thermal cycle).
• Step B — Temperature sweep
  Measure bias and stability across temperature; drift-vs-temp curves separate sensor physics from incidental transient triggers.
• Step C — Noise density and bias drift curves
  Static placement and controlled motion tests reveal noise floors; compare across system states (display/radio/charging).
• Step D — Mechanical stress checks
  Press/flex and assembly-condition comparisons identify offset steps caused by packaging and mechanical stack-up.
• Step E — Bus evidence (supporting)
  Track I²C/SPI error/retry bursts and timing anomalies; align with false triggers and unstable readings.

Thermal & Mechanical Constraints: Heat Paths That Rewrite “Best Specs”

In smartphones, “performance” is often rewritten by thermal paths and mechanical stack-ups: PA output is derated, SoC clocks throttle, battery voltage droops under load, and touch/sensors drift with thermal gradients. The practical way to debug is to treat heat as an event chain—hotspot → heat path → sensor points → policy triggers → user-visible symptoms— then close the loop with time-aligned evidence.

• Hotspot (where heat is generated)
  SoC, PA/FEM, battery, display/backlight power blocks.
• Heat path (where heat flows)
  Die → interface → midframe/VC/graphite → housing; local bottlenecks create sharp thresholds.
• Sensor points (what the device “sees”)
  Die temp, skin temp, battery temp; different sensors trigger different limits.
• Policy triggers (what changes)
  DVFS/thermal throttling, PA backoff/derating, charge current limits, brightness limits.
• Symptoms (what users notice)
  FPS drop, throughput drop, call drops, random reboot, touch drift/ghost touches.

• A) PA/FEM derating
  Symptoms: throughput drops or call drops during sustained uplink. Proof: PA backoff/derating events align with PA-side temperature rise and link-quality changes.
• B) SoC throttling
  Symptoms: frame drops, UI lag, camera pipeline instability. Proof: DVFS/thermal throttling logs align with performance drops at threshold crossings.
• C) Battery IR + VBAT droop under load
  Symptoms: “random reboot” or sudden power cut under bursts. Proof: VBAT droop aligns with load state + reset reason; battery temperature and charging state explain margin loss.
• D) Touch/sensor drift from thermal gradients
  Symptoms: touch drift/ghost touch, sensor false triggers. Proof: touch baseline/noise map or sensor bias drift changes predictably with temperature sweep and device state.

• SoC path: hotspot near compute blocks; bottlenecks at interfaces and local spreading often create step-like throttling thresholds.
• PA/FEM path: localized edge heating can force PA backoff even when SoC is stable.
• Battery path: temperature and internal resistance reduce transient headroom, increasing droop risk during bursts.
• Display/backlight path: sustained brightness and switching heat can contribute to regional gradients that bias touch/sensors.

• Multiple sensors, multiple triggers: die/skin/battery thresholds can activate different limiters (DVFS, PA backoff, charge limit).
• Threshold crossings create steps: once a threshold is crossed, performance can drop abruptly even if temperature changes slowly.
• Debug goal: identify which sensor point is authoritative for the observed limiter.

• Step A — Locate hotspots
  Use thermal imaging or targeted thermocouples to find where temperature rises fastest under the suspect workload.
• Step B — Mark load-state transitions
  Label state changes (uplink burst, camera start, gaming load, charging) and capture current/voltage where possible.
• Step C — Collect limiter logs
  DVFS/thermal throttling, PA derating/backoff, charge current limits, brightness limits—then time-align to symptoms.
• Step D — Confirm with controlled deltas
  Change only one factor (case airflow, brightness, uplink duty, charge state) and verify the threshold crossing moves predictably.

Robustness: ESD/EFT/Surge-like Events on User Interfaces (Device-Side)

Real-world “electric shocks,” plug events, and static discharges commonly enter through user interfaces: touch edges, USB-C shells and pins, buttons, and audio paths. The outcome is often not immediate failure, but drift and instability—touch anomalies, port dropouts, audio noise, or resets. The key is to map entry points and return paths, then verify protection effectiveness without creating new problems from TVS/RC loading.

• Touch edge / frame
  Direct injection into touch reference and scan window; after-event baseline drift is common.
• USB-C (shell + pins)
  Shell and CC/SBU/high-speed lines can see injection during plug/unplug and discharge.
• Buttons / side keys
  Coupling through metal frames and local traces into sensitive domains and references.
• Audio path
  Ground/reference disturbance and return current spikes show up as pops/noise or degraded SNR.

• Touch abnormal (ghost touch / drift / dead zones)
  Coupling path: touch reference + scan window interference. Capture: raw/baseline/noise map before/after the event and compare drift deltas.
• USB port dropouts / failure to enumerate
  Coupling path: CC/SBU/state machine upset or PHY stress. Capture: reconnect/retry logs, CC state changes, and event reproduction conditions.
• Audio noise / pop / distortion
  Coupling path: ground/reference shift into codec/amp. Capture: transient waveform, FFT before/after, and ground-return inspection.
• Reset / reboot
  Coupling path: ground bounce or rail disturbance trips reset/fault. Capture: reset reason and rail droop/ripple aligned to the event.

• Return loop first: a TVS placed far away with a long loop often performs worse than a near-entry clamp with a short return.
• Keep the sensitive reference clean: touch/audio references are vulnerable to shared return impedance and local ground splits.
• Verify by repeatability: if the same point and posture reproduces the fault, the coupling path is physical and fixable.

• TVS capacitance loading
  Can increase touch noise, degrade fast edges, or reduce margin—especially near high-speed or sensitive references.
• RC / series-R timing shift
  Can change edge timing and windows, creating new failure modes (false triggers, distortion, link retries).
• Proof technique
  Compare “before/after” noise maps, retry counts, or distortion metrics while holding the same ESD point and conditions constant.

• Step A — Choose and document points
  Touch edge, USB shell, buttons, and audio reference points; keep the point set stable between runs.
• Step B — Lock the reproduction conditions
  Charging on/off, cable type, brightness/refresh state, radio activity, hand posture and grounding conditions.
• Step C — Capture the right artifacts
  Touch raw/baseline/noise map; USB retry/reconnect logs; audio transient and FFT; reset reason and rail evidence.
• Step D — Inspect the protection loop
  TVS/RC location, shortest return, ground continuity; avoid long loops and split references near sensitive blocks.
• Step E — Validate trade-offs
  Confirm that added TVS/RC does not create new instability via capacitance/timing/noise side effects.

Validation & Bring-up Plan: What to Measure First (PI/RF/HMI/Sensors)

A smartphone bring-up plan must be executable and regression-friendly. Use a strict order that respects coupling: stabilize power/reset first, validate RF next, then lock display/touch, and finally close sensor drift and false-trigger risks. Every step must define pass/fail evidence that can be captured, compared, and replayed.